Update gain normalization is employed in adaptive filters to control weighting of the filter impulse response updates in order to converge properly to a desired impulse response. Singing, i.e., oscillating, of the filter is overcome by adjusting the update gain when an incoming signal power estimate used to normalize the gain exceeds a prescribed threshold value. In one example, the normalized gain is adjusted to be a fixed value for power estimate values which exceed the threshold. In accordance with another aspect of the invention, a single normalized gain value is used to adjust the update gain in two adaptive filters employed as echo cancelers in a bidirectional voice frequency repeater.

An estimation apparatus predicts filter coefficients for an adaptive filter, the response of which emulates the signal transmission characteristics of a known signal. The response thereto is sent to a signal transmission system of unknown characteristics, enabling execution of calculations without invalidating coefficient updating, even when there is a limit on the word length for processings. To achieve this, a sum of products calculation unit calculates the sum of products of the residual difference in response and the signal which is sent to the signal transmission system. A sum of squares calculating unit calculates the sum of the squares of the signal sent to the signal transmission system over a prescribed period of time. An updating amount calculation unit calculates the filter coefficient updating amounts from the ratio of the results from the sum of the products calculating unit to the results of the sum of the squares calculating unit. Filter coefficients are updated using the coefficient updating amounts calculated by the updating amount calculation unit.

A transmission method for digital data which is adapted to transmit a predictive residual at every sample and a predictive coefficient and a normalization gain at every data block to thereby provide highly efficient coding and data transmission. An adaptive quantization method suitable for use in the above-mentioned digital data transmission, forms samples into a block and a predetermined multiplication is effected on the sample of each of the blocks to thereby improve the accuracy of the predictive coefficients. A stable operation may be ensured in the adaptive quantizing method by cancelling the subsequent operation when the auto-correlation of input data is not recognized.

A active noise control system having a digital signal processor programmed to function as adaptive digital filter with an output saturation control is described. The programmed adaptive filter has a specified maximum value for its output signal, above which the output signal becomes clipped and a condition commonly known as filter output saturation occurs. According to the described system, the filter output signal value is monitored to detect the occurrence of filter output saturation. When output saturation is detected, the normal adaptation of the filter is modified to reduce the value of the filter output signal to prevent the further occurrence of filter output saturation. Based upon the detection of filter output saturation, a power reference level is established to estimate the maximum average power that the output signal can contain without causing saturation. The adaptation of filter weighting coefficients is then scaled in accordance with the difference between the established power reference level and a computed average power associated with the filter output signal to prevent the occurrence of filter output saturation.

An electronic compensation network which allows the analytic and empirical realization of the compensation needed to optimize a system response. The electronic compensation network includes a switched capacitor active filter network; means for selectively changing filter characteristics; and, means for indicating the characteristics of the compensator which produces an optimized system response.